►
From YouTube: BTS OpenShift Common - Vlad Shlosberg (Foqal)
Description
Behind the Scenes at OpenShift Commons with Michael Hausenblas
Interview with Vlad Shlosberg of Foqal
A
B
A
B
A
B
You
see
my
demo
slack
screen,
thingy
yeah,
yeah,
perfect
cool,
so
what
basically
we'll
focal
does
is
it
will
automatically
use
AI
to
detect
questions
and
automatically
find
you
an
answer
to
your
toughest
questions.
So,
let's
imagine
I,
already
kind
of
wrote
one
of
the
questions
but
like
let's
imagine,
somebody
comes
in
and
goes
I
how
do
I
install
Red
Hat,
though
without
any
kind
of
prompting,
without
any
kind
of
like
listed
Lee?
B
Invoking
the
thing
vocal
will
detect
that
you're
asking
a
question
find
the
most
relevant
answer
from
within
all
of
your
internal
sources
and
tries
to
give
you
the
best
answers
that
it
can
find
all
right.
So
in
this
case,
if
you
look
it
will,
you
know,
found
one
and
one
answer
from
a
previous
lab
conversation.
B
You
still
have
a
question,
or
is
it
not
helpful,
and
the
idea
here
is
that
if
you
click
like
ants
will
helpful
and
answer
a
question
then
because
this
is
only
visible
to
you-
we
want
to
notify
the
rest
of
the
channel
that
your
question
was
already
answered,
so
they
don't
have
to
think
about
it.
So,
basically,
this
is
the
kind
of
like
the
most
easiest
way
that
people
interact
is
that
they
come
in
regular
channel.
They
don't
even
know
that
this
thing
exists.
B
They
just
go,
ask
a
question
and
automatically
it
gets
answered,
and
nobody
else
has
to
think
about
it.
Nobody
else
has
to
be
distracted
from
their
day.
Jobs
of
doing
this,
though
so
there's
a
couple
of
places,
there's
a
couple
of
things
where
these
questions
come
from.
One
thing,
you'll
see,
is
again
it's
confluence
of
previous
slack
conversations.
So
let
me
show
you
how
that
previous
slack
conversation,
things
work,
launch
my
demo
script
and
something
else.
I
guess.
B
So
imagine
you
have
two
people
talking.
You
have
Michael
NK
and
Michael
asks
K
a
question
like
hey.
How
do
I
create
a
JavaScript
array?
You
know
K
helps
Michael
Michael's
happy
right,
so
at
that
point
focal
not
only
detects
the
question
that
Michael
is
asking,
but
it's
also
in
detecting
the
answer
that
K
provides
to
Michael
and
it
will
send
a
message
in
this
case
is
saying
to
me
for
demo
purposes,
but
I'll
actually
send
a
message
to
K,
say:
okay,
we
found
this
is
something
useful.
B
We
believe
this
is
something
that
could
be
useful
in
the
future.
Let
us
store
it
for
future
right
and
she
can
click
ignore
and
it'll
go
away,
because
maybe
it's
you
know
private
or
something
like
that.
You
can
edit
the
question
answer
to
provide
more
context
or
you
know
whatever
it
is,
and
then
she
can
finally
click
store
and
as
soon
as
she
does
that
it
becomes
available
as
a
potential
like
automatic
answer
to
people
asking
kind
of
similar
question
in
the
future.
B
Yeah
exactly
exactly
so,
and
this
is
why
it's
very
much
designed
to
you
know
all
the
messages
are
only
visible
to
you.
You
can
invoke
it
directly
by
just
like
using
the
slash,
vocal
command
or
something,
but
you
know
the
whole
goal
here
is
that
we
don't
expect
people
to
necessarily
always
do
that.
We
expect
people
are
gonna.
Just
not
know
this
thing
exists,
come
in,
ask
a
question
and
we
don't
want
to
annoy
the
rest
of
the
channel,
wouldn't
bother
us
or
channels.
B
So
what
kind
of
messages
directly
to
the
person
right,
even
in
our
most
successful
installs,
a
lot
of
the
people
don't
even
know
that
it's
kind
of
there,
because
unless
they
ask
the
question
that
we
can
interact
with
unless
we
ask
unless
somebody
has
a
question
that
we
can
answer
for
you,
many
people
don't
even
know
that
it
exists
until
they
get
value
out
of
it
and
so
they're.
Getting
like
something.
That's
cool,
gotcha.
B
A
Right
right,
so
you
mentioned
a
couple
of
sources
that
could
be
external
internal
I.
Believe
you
know,
if
you
think
about
the
deployment
behind
the
firewall
Enterprise
setup.
How
do
you
like
where
and
how
do
these
sources
come
into
place?
That's
something
where
you
know
you
can
configure
in
a
per
channel
basis
or
or
how
do
you
actually
make
focal,
like
you
know,
pay
attention
to
a
certain
knowledge
bases
order,
gotcha.
B
So
so,
there's
a
administrative
component
to
this
entire
thing.
This
is
basically
just
one
example,
but
the
idea
here
is
that
you
can
go
in
and
just
kind
of
add
yourself
to
some
integrations
and
generally.
What
it
is
is
that
you,
you
know
you
go
to
this
like
admin
site.
You
currently
were
offering
these
three
external
ones
that
you
can
self
configure
and
then
there's
one
that
I
can
configure
for
you.
B
But
the
idea
is
that
we
can
currently
just
crawl
your
kind
of
Zendesk
and
if
lassi
and
cloud
products
we
can
crawl
kind
of
any
kind
of
public
website.
So
in
this
case
you
know
this
is
my
demo
one.
So
it's
like,
like
it's
scraping
some
like
how
to's
and
that
kind
of
stuff-
and
it's
just
pushing
it
in
there
and
then
we're
currently
working
on
Google
Docs,
we're
currently
so
with
Atlassian.
We
were
working.
We
currently
supporting
Giro
we're
currently
working
on
confluence,
so
we're
kind
of
building
up.
A
And
I
remember
from
like,
like
how
I
stumbled
upon
the
course
essentially
through
the
the
quiddity
slack
right
and
I.
Remember
there
was
the
the
stack
overlook,
but
it's
like
a
low-flow
integration
right
at
that
desk.
So
any
any
kind
of
these
Q&A
tides,
I,
guess
that
makes
a
sense.
We
already
have
this
kind
of
like
here's,
the
question
that
and
then
you
have
potential
ranked
answers
for
that
yeah.
B
And
actually
one
of
our
some
of
our
best
answers
actually
comes
from
not
necessarily
the
structured
question-and-answer
formats,
but
it's
actually
from
like
Docs,
so
some
of
our
best
Doc's
come
from
Earth.
All
of
our
best
answers
actually
come
from
going
to
these
open
source
like
Doc,
you
know
pages
like
these
kind
of
things,
and
we
do
you
scrape
them
or
you
know,
download
them,
maybe
scrape
isn't
Isis
for
the
safe,
we
borrow
them
and
we
we
take
them
apart
into
smaller
chunks,
because
the
idea
here
is
like
right
a
lot
of
times.
B
B
It
up
into
smaller
chunks
and
then,
when
somebody
has
a
question,
we
send
them
the
exact
chunk
with
that
text,
kind
of
reformat
it
as
a
slack
message
and
then
we'll,
of
course
give
them
like
a
reference
to
exactly
in
the
dock
where
that
thing
is-
and
that's
actually
been
so
far
like
the
best
like,
like
looking
at
numbers
on
one
of
our
biggest
customers,
be
the
best
answers
like
so
like
I
just
pulled
the
numbers.
Now
in
the
last
couple
of
few
months,
54
of
the
answers
came,
they
were
helpful.
B
They
were
marked
as
like
something
that's
actually
useful
to
people
54
of
the
times
it
came
from
Stack
Overflow
260
came
from
previous
conversations
that
were
kind
of
required,
like
in
the
demo,
showed,
and
then
286
came
from
previous
Docs.
So,
interestingly
enough,
a
lot
of
the
answers
actually
do
come
from
just
you
know,
you
have
some
Doc's,
we
go
and
basically
download
the
docs.
You
know
slightly
reformat
them
and
make
them
available,
and
you
know
which
kind
of
begs
the
question
of.
Like
you
know.
Did
you
read
the
effing
manual
write.
A
Guess
it's
all
about
the
context,
but
if
you
are
able
to
actually
pull
the
exact
you
know
paragraph
or
whatever
section
someone
is
looking
for
it's
kind
of
like
you
know,
this
is
a
friendlier
version
of
RTFM
right,
yeah
yeah,
it's
yeah,
yeah
I,
don't
have
time
to
read
like
you
know,
500
pages,
but
I
I
actually
appreciate
this
one,
this
one
section
that
actually
answers
or
contains
the
answer
to
my
question:
awesome
cool
any
more
on
the
technical
end.
Well,.
B
And
actually
so,
though,
I
wanted
to
mention
that
you
brought
up
docs.
The
one
interesting
thing
about
Doc's
is
actually
were
working
with,
like
the
docs
communities,
we're
actually
I'm
gonna
install
this
thing
on
a
one
of
these
like
big
slide
channels
for
tech
writers.
Soon
we're
working
with,
like
you
know,
in
tech
writers
in
general
and,
like
you
know,
people
that
create
these
dogs
and
the
interesting
thing
that
this
is
providing
for
them
is
that
it's
giving
a
lens
into
exactly
who
the
reader
is
in
the
past
people
say
like
ok,
Who
am
I.
B
Writing
these
docs
like
who's
my
target
reader
right
and
you
can
kind
of
guess
what
the
profile
is,
but
here
you're
actually
getting
a
very,
very
specific
view.
What
are
the
questions
that
people
are
asking
and
who
are
they?
What
who
are
they?
What
are
they
trying
to
get
at
and
and
even
from
like
the
micro
level?
It's
like
just
to
see
how
what
question
is
the
most
being
asked
is
kind
of
like
indication
of
what
might
we?
A
From
a
related
note
like
the
well
I
am
a
big
fan
of
the
user
of
slag.
Obviously
the
integration
with
with
slag
as
a
kind
of
front-end,
it's
the
interaction,
art,
it's
kind
of
random
right
I
mean
this
just
happens
to
be
like
yeah,
a
nice
starting
point,
because
that's
where
a
lot
of
people
are
in
the
community
but
I
mean
nothing
holds
you
back.
You
know
there
are
these
these
in
in
web,
app
with
website
chads2
pop
up.
B
Yeah
definitely
I
mean
so
again.
Yes,
slack,
as
you
said,
is
just
the
starting
point.
It's
just
the
place
where
we
know
that
there's
a
lot
of
people
today,
we
know
for
the
open
source
world
there's
a
lot
of
people
using
slack
and
I
mean
there's
look
kind
of
like
the
movement
from
IRC
I.
Think
a
lot
of
people
are
getting
out
of
IRC
and
moving
into
slack
also
in
the
corporate
world.
B
A
lot
of
people
are
using
slack
right
and
there's
just
this
kind
of
argue
this
battle,
that's
brewing
between
slack
and
Microsoft's
moves
where
we
don't
have
to
choose
a
side.
We
can
actually
just
support
both
and
then
so
other
than
my
you
know,
Microsoft
teams,
the
next
thing
that
you
might
actually
even
think
about
as,
like
you
know,
you're
thinking
of
like
oh
imagine,
you're
in
word,
and
somebody
props
up
something.
Well
then
you're
thinking.
Well,
then,
oh,
that's
quickly!
That's
annoying
right!
Well,
the
thing.
A
A
B
The
same
time,
if
you
look
at
Google
Docs,
they
actually
added
this
Explorer
feature
which,
like
in
the
bottom
right-hand
of
every
you,
know
Docs
and
sheets
and
presentation
and
stuff.
They
have
this
feature
that,
like
basically
goes
and
analyze
your
data
and
gives
you
additional
insights
on
your
data
right.
So
up
so
in
a
way,
that's
like
a
more
modern
version
of
Clippy,
which
I
imagine
this
would
be
something
as
well
right.
Is
that
imagine
you're
writing
a
doc
and
not
necessarily
that
we're
popping
off
with
saying
hey.
B
Are
you
trying
to
write
a
letter?
Let
us
help
you
write
it
right
or
so,
like
they
explore
feature
that
basically
analyze
your
information
and
it
Eliza's
your
actual,
like
Doc
or
analyzes,
your
like,
let's
say
sheet
and
says:
oh
here's
some
averages,
here's
like
a
graph
of
this
kind
of
day
and
then
stuff
it
was
basically
it's
giving
you
an
additional
insights
based
on
your
existing
data.
So
here
is
it's
kind
of
like
what
we're
thinking
about
is
the
same
direction
of
like
well.
B
What,
if
we're,
providing
additional
insights
on
you're,
not
just
your
current
doc,
that
you're
looking
at
but
all
of
your
company
data
and
helping
you
make
either
you
know,
make
making
helping
you
make
business
decisions
on
your
data
in
a
very,
very
quick
rate,
because
what
we're
gonna
do
is
we're.
Gonna,
recommend
small
tidbits
of
information
that
might
that
are
supposed
to
be
really
really
relevant
to
what
you're.
Looking
at
what
you're
doing
today,
what
you're
doing
right
now.
A
Let's
switch
gears
a
little
bit
and
talk
a
little
bit
about
the
machinery.
Part
I'm,
not
expecting
you
to
reveal
your
trade
secrets
or
whatever,
but
I
gather
or
I
understand
that
it
means
it's
a
little
bit
more
than
just
bring
it
matching
like.
This
is
a
question
because
it
starts
with
what
is
or
how
and
I
or
whatever
you're
doing
a
little
bit
more.
There
so
kind
of
understanding
where
people
like
what
people
are
trying
to
do,
and
not
just
matching
question
marks
or
whatever.
B
Yeah,
actually,
what
you
just
described
is
it
would
be
in
arc
in
most
of
our
cases,
a
bad
way
to
detect
a
lot
of
times.
What
happens
is
somebody
comes
in
and
says
hi
I'm
having
this
problem
and
that's
the
question
and
then
the
response
would
be
something
like
have
you
tried
block
question
mark,
don't
know,
so
it's
like
very,
very
reverse
in
that
situation,
so
yeah
I
mean
the
machine.
Learning
has
to
be
able
to
detect
that
kind
of
stuff
and
it's
basically
trained
along.
You
know
straight
along,
like
this
kind
of
conversational.
B
You
know
text
to
say
like
what
looks
like
a
question.
What
looks
like
an
answer?
How
do
we
detect
one
to
the
other?
The
other
interesting
part
about
the
AI?
Is
that
like
it's?
Not
just
or
you
know,
machine
learning,
I
don't
like
to
really
call
it
AI.
The
interesting
thing
about
that
is
that
it's
not
only
you
know,
just
detecting
the
question
answer.
B
One
of
the
interesting
one
of
the
hard
parts
is
actually
that
in
a
in
a
slack
channel,
there
might
be
multiple
conversations
going
on
at
the
same
time,
and
you
know
you
as
a
person
you
know
who
you're
talking
to
so
you
kind
of
you
know
read
it
anyway
and
you
can
maybe
pay
attention
to
percent
so,
like
whatever
the
conversation.
The
other
conversation
is
in
the
channel,
but
you
need
to
also
you're
mainly
paying
attention
to
the
person
that
you're
talking
to
so
the
bot
has
to
kind
of
do
that
too.
B
B
So
there's
some
interesting
things
on
that
ground.
There's
some
interesting
zone,
questions
and
answer
detection,
and
then
you
know
and
like
and
then
on
the
other
side
of
like
once
we
find
a
question
and
answer:
how
do
we
find
it?
That's
the
most
useful
answer:
how
do
you
convert
like
something
that
people
are
not
expecting
to
be
used?
This
question
right
when
somebody
just
comes
in
and
says:
hi
I
have
a
problem.
How
do
you
convert
that
into
something
that
we
can
actually
like
answer
a
question
on
right.
A
B
Maybe
using
you
know,
how
do
we
use
more
microsecond
ologies
and
how
do
we
create
some
integrations
and
so
on?
There
was
like
these
big
gaps:
I'm
just
like
people
trying
to
do
some
work
on
one
side,
people
trying
to
support
that
work
on
the
other
side
and
there's
these
big
holes
in
the
middle
of
just
collaboration
and
communication,
where,
like
somebody,
would
say
something,
and
then
that
message
only
gets
to
the
appropriate
party.
B
Maybe
two,
three
weeks
later,
when
they're
already
been
churning
away
on
this
old
information,
so
I've
seen
just
problems
like
this
at
LinkedIn
and
the
previous
companies,
and
just
and
from
there
I
kind
of
left
decided
to
go
figure
out
how
to
solve
some
of
these
problems.
Talk
to
hundreds
of
people
like
something
doing
this
for
about
a
year.
Talks
about
you
know
a
hundred
people
in
the
beginning,
just
before
even
formalizing.
B
This
idea
started
building
this
started
kind
of
getting
early
feedback
and
really
only
launched
this
product,
as
as
it
kind
like
as
a
the
first
version
of
it
around
March
and
it's
obviously
been
iterating
ever
since,
but
I'm
sorry
June
in
March,
I
kind
of
had
the
concept
and
started
building.
So
that's
the
students
about
four
months
been
iterating
gods
who
were
installed
in
about
seven
customers
or
seven
places.
So
far,
it's
doing
well,
it's
gonna!
A
So
I
can
I
can
just
from
my
own
experience,
say
to
me
and
and
like
also
someone
who
is
trying
to
answer
and
helping
people,
and
it's
like
super
valuable
because
a
lot
of
these
things,
especially
if
it's
repetitive,
like
the
500th
person,
asked
about
pad
away.
You
know
expose
that
service
like
well.
It
is
you
know
you
come
into,
like
you
know,
I
give
em.
This
is
really
like.
A
It's
really
really
awesome
great
UX
and
yeah
I'm
super
big
fan,
maybe
but
like
quick
like
detour
or
like
changing
the
angle,
a
bit
in
terms
of
with
understand
and
it
stood
like
what
what
is
going
on
but
happen
where
they
come
from,
and
why
you're
doing
that?
What
are
the
next
steps?
Where
were
you
heading?
What's
your
whole
idea
with
that?
What
do
you
want
to
okay.
B
That's
a
very
interesting
question,
so
I'm
looking
at
so
obviously
growing
it
some
more
spaces
and
using
that
to
improve
our
data.
One
of
the
things
I'm
looking
at
right
now
is
just
to
you
know:
I've
been
doing
a
you
know
like
I'm,
a
a
I
I'm,
not
really
an
AI
guy
I'm,
not
really
a
data
scientist,
so
I'm
trying
to
find
help
in
that
regard.
So
one
of
the
things
I'm
doing
kind
of
the
short
term
was
trying
to
figure
out
how
to
get.
You
know
they
I
better.
B
Just
you
know
10x
trying
to
improve
that
those
fields
at
the
same
time,
I'm,
basically
growing
it
out
to
more
communities
and
stuff
kind
of
early
at
a
slightly
slower
pace,
just
trying
to
make
sure
that
we're
hand-holding
everybody
through
make
sure
that
we're
providing
value
from
the
beginnings
that
people
are
seeing
with
with
the
you
know
what
this
can,
how
they
can
benefit
from
this
just
from
the
beginning,
and
we
don't
want
to
grow
too
too
fast
or
to
make
sure
everyone's
happy
and
then
over
the
next
few
months.
It's
going
to
be
about.
B
You
know
trying
to
get
this
on
more
places,
trying
to
get
better
data
using
that
trying
to
you
know
scale
this
up
to
just
you
know,
kind
of,
like
maybe
maybe
not
as
many
users,
but
try
to
scale
it
up
to
provide
as
much
value
as
we
possibly
can
to
whether
it's
all
swag
communities,
and
then
you
know
kind
of
from
there.
Other
communities-
and
you
know,
try
to
see
some
how
we
like.
So
it's
already
formed
as
a
as
a
as
a
corporation.
B
A
Maybe
last
question
from
my
end
well
like:
if
you
have
anything
to
share
or
show
please
please
do,
but
one
thing
that
I
would
be
interested
in,
especially
you
know
people
out
there
when
when
this
goes
live,
you
were
like
awesome
like:
where
do
they
start
right?
Do
they
send
you
an
email
or
have
you
maybe
got
website,
but
how
do
we
get
it?
Can
they
download
it
somewhere?
Do
they
need
to
ask
you
to
install
it
where
what,
if
you
go,
where
do
they
go.
B
B
Talk
to
me
and
then
we'll
kind
of
go
create
an
account
for
you,
one
of
the
things
I'm
trying
to
work
on
over
the
next
like
month
or
so
is
to
just
really
streamline
that
process
where
essentially
be
able
to
go
and
focal
that
IO
and
just
install
it
yourself,
and
you
don't
have
to
kind
of
go
through
me
or
anything
like
that,
so
there's
kind
of
two
sites
on
there
right
now.
One
of
them
is
just
for
the
more
corporate
side
as
ika.
B
If
you
want
to
go
install
this
at
your
company,
go
right
ahead.
Here's
the
signup
screen,
the
other
side
is
the
kind
of
like
it's.
You
know:
slash
OSS
likes
open-source
software,
it's
more
designed
not
necessarily
for
open
source
software,
but
it's
designed
for
open
communities
where,
if
anybody's
allowed,
if
there's
no,
you
know,
there's
no
requirement
for
you
to
either
pay
for
that
pay
to
join
that
slack
or
there's
no
requirement
for
you
to
be
a
company
to
join
that
slack.
So
it's
and
I'm
open
for
those
kind
of
teams.
B
B
A
I
I,
gotta,
you're,
very
responsive
and
hopefully,
at
some
point
in
time.
You
you
won't
have
to
try
to
be
responsive
anymore
because
you
were
you're
busy,
counting
all
the
dollar
bills
and
whatnot
and
and,
like
I,
said
I'm.
A
big
fan
and
I
really
really
hope
that
you
know
for
you
this.
This
works
out,
because
I
think
this
is
like
to
meter
the
big
next
step
and
necessary
in
this
this
whole
little,
but
but
world
or
whatever,
and
thinking
of
having
that
on
the
mobile
phone,
maybe
on
some
other
environment.
A
B
You
know
it's
not
gonna
be
thank
you,
for
you
know
getting
value
out
of
the
product
and
enjoying
it
and,
and
you
know
any
kind
of-
hopefully
you
you
know
you're,
finding
you
more
value
and
hopefully
becomes
even
better
over
time.
Yeah
I
mean
I.
Guess
the
only
take-home
messages
hey
go
install
it.
Let
me
know
how
you
feel
tell
me
you
know
I'm
early
on
everyone
as
much
feedback
as
humanly
possible,
so
it
the
best
feedback.
I'm
getting
is
just
like
you.